Systems and methods for classifying or selecting images based on image segmentation

Abstract

The present disclosure relates to classifying and/or selecting images based on image segmentation. A classification system for classifying images includes one or more processors and at least one memory storing machine executable instructions. When the instructions are executed by the one or more processors, they cause the classification system to: access image segmentation scores for pixels of an image, and classify the entire image based on the image segmentation scores for the pixels of the image. The image segmentation scores for the pixels of the image are provided by an image segmentation system based on the image, and each of the image segmentation scores correspond to at least one pixel of the pixels of the image.

1 - 20 . (canceled) 
     
     
         21 . A classification system for classifying images, comprising:
 one or more processors; and   at least one memory storing machine executable instructions which, when executed by the one or more processors, cause the classification system to:
 access image segmentation scores for pixels of an image, the image segmentation scores provided by an image segmentation system based on the image, each of the image segmentation scores corresponding to at least one pixel of the pixels of the image; 
 classify the entire image based on the image segmentation scores for the pixels of the image; 
 identify a cluster of pixels of the image corresponding to a cluster of highest image segmentation scores among the image segmentation scores for the pixels of the image; and 
 determine an average score of the cluster of highest image segmentation scores. 
   
     
     
         22 . The classification system according to  claim 21 , wherein the instructions, when executed by the one or more processors, further cause the classification system to transform the image segmentation scores for the pixels of the image to provide at least one image classification score,
 wherein classifying the entire image includes classifying the image based on at least the at least one image classification score.   
     
     
         23 . The classification system according to  claim 22 , wherein in transforming the image segmentation scores, the instructions, when executed by the one or more processors, cause the classification system to perform at least one of an inference of a machine learning system or a non-machine learning transformation operation. 
     
     
         24 . The classification system according to  claim 23 , wherein the non-machine learning transformation operation includes determining a maximum score among the image segmentation scores for the pixels of the image,
 wherein classifying the entire image includes classifying the entire image based on the maximum score.   
     
     
         25 . The classification system according to  claim 23 , wherein the non-machine learning transformation operation includes determining at least one of:
 an average score of a predetermined number of highest image segmentation scores among the image segmentation scores for the pixels of the image; or   a count of the image segmentation scores for the pixels of the image having a value above a threshold, wherein classifying the entire image includes classifying the entire image based on at least one of the average score or the count.   
     
     
         26 . The classification system according to  claim 23 , wherein identifying the cluster of pixels and determining the average score of the cluster of highest image segmentation scores is part of the non-machine learning transformation operation, wherein classifying the entire image includes classifying the entire image based on the average score of the cluster. 
     
     
         27 . The classification system according to  claim 21 , wherein the image segmentation scores for the pixels of the image include scores indicating whether a pixel is a background pixel or a pixel of interest, wherein the instructions, when executed by the one or more processors, cause the classification system to further perform at least one of:
 determining a shape of pixels indicated to be pixels of interest based on the image segmentation scores; or   determining a distribution of pixels indicated to be the pixels of interest based on the image segmentation scores, wherein classifying the entire image includes classifying the entire image based on at least one of the determined shape or the determined distribution.   
     
     
         28 . The classification system according to  claim 21 , wherein the instructions, when executed by the one or more processors, further cause the classification system to input the image to a deep learning neural network to generate the image segmentation scores. 
     
     
         29 . The classification system according to  claim 21 , wherein each score of the image segmentation scores corresponds to one pixel of the pixels of the image. 
     
     
         30 . A classification system for classifying images, comprising:
 one or more processors; and   at least one memory storing machine executable instructions which, when executed by the one or more processors, cause the classification system to:
 access image segmentation scores for pixels of an image, the image segmentation scores provided by an image segmentation system based on the image, each of the image segmentation scores corresponding to at least one pixel of the pixels of the image; 
 classify the entire image based on the image segmentation scores for the pixels of the image; and 
 at least one of:
 determine a shape of pixels indicated to be pixels of interest based on the image segmentation scores; or 
 determine a distribution of pixels indicated to be the pixels of interest based on the image segmentation scores. 
 
   
     
     
         31 . The classification system according to  claim 30 , wherein the instructions, when executed by the one or more processors, further cause the classification system to transform the image segmentation scores for the pixels of the image to provide at least one image classification score,
 wherein classifying the entire image includes classifying the image based on at least the at least one image classification score.   
     
     
         32 . The classification system according to  claim 31 , wherein in transforming the image segmentation scores, the instructions, when executed by the one or more processors, cause the classification system to perform at least one of an inference of a machine learning system or a non-machine learning transformation operation. 
     
     
         33 . The classification system according to  claim 32 , wherein the non-machine learning transformation operation includes determining a maximum score among the image segmentation scores for the pixels of the image,
 wherein classifying the entire image includes classifying the entire image based on the maximum score.   
     
     
         34 . The classification system according to  claim 32 , wherein the non-machine learning transformation operation includes determining at least one of:
 an average score of a predetermined number of highest image segmentation scores among the image segmentation scores for the pixels of the image; or   a count of the image segmentation scores for the pixels of the image having a value above a threshold,   wherein classifying the entire image includes classifying the entire image based on at least one of the average score or the count.   
     
     
         35 . The classification system according to  claim 32 , wherein non-machine learning transformation operation includes:
 identifying a cluster of pixels of the image corresponding to a cluster of highest image segmentation scores among the image segmentation scores for the pixels of the image; and   determining an average score of the cluster of highest image segmentation scores,   wherein classifying the entire image comprises classifying the entire image based on the average score of the cluster.   
     
     
         36 . The classification system according to  claim 30 , wherein classifying the entire image includes classifying the entire image based on at least one of the determined shape or the determined distribution. 
     
     
         37 . The classification system according to  claim 30 , wherein the instructions, when executed by the one or more processors, further cause the classification system to input the image to a deep learning neural network to generate the image segmentation scores. 
     
     
         38 . The classification system according to  claim 30 , wherein each score of the image segmentation scores corresponds to one pixel of the pixels of the image. 
     
     
         39 . A classification method for classifying images, comprising:
 accessing image segmentation scores for pixels of an image, the image segmentation scores provided by an image segmentation system based on the image, each of the image segmentation scores corresponding to at least one pixel of the pixels of the image;   classifying the entire image based on the image segmentation scores for the pixels of the image;   identifying a cluster of pixels of the image corresponding to a cluster of highest image segmentation scores among the image segmentation scores for the pixels of the image; and   determining an average score of the cluster of highest image segmentation scores.   
     
     
         40 . The classification method according to  claim 39 , wherein classifying the entire image includes classifying the entire image based on the average score of the cluster.